三維濾波器應用在實際頭殼上之腦電位重建

Abstract

本論文之主要研究目的是探討三維(3D)濾波器，對於實際頭殼模型之 腦電位分布處理的效能; 其中，腦電位分布的處理受限於有限的錄製點數 目。先前的研究中，以球殼模型做模擬實驗，結果顯示三維濾波器可達 到4NN法(four-nearest-neighbor method)和spherical splines等廣為使 用之方法的效果。 本論文中，我們首先使用理想球殼頭殼模型來分 析(1)錄製點數目多寡及(2)三維空間的取樣解析度對重建後腦電位分布的 影響。在此部份，我們採用上述三種方法並比較其效能。其後，我們使用 三維濾波器法與4NN法來處理實際頭殼模型之腦電位重建。各種方法之結 果以下述兩種法則來評估其效能：(1)腦電位分布的理論值與重建值得比 較，(2)真實與重新定位之病變源的比較。 The purpose of this research is to investigate the feasibility of applyingthe three-dimensional(3D) filtering technique to manipulating the brain potential distribution on a realistic head model; based on a limited number ofEEG( electroencephalograph) recording electrodes. In previous study, the 3D filtering method based on a spherical model was shown to have the performancecomparable to that of the 4NN(four-nearest- neighbor) and the spherical splinesmethods. In this thesis, we first utilize the spherical model to analyze the effect of the number of recording electrodes and the sampling resolution of the 3D space on the interpolated brain potential mapping. The three methods areemployed for comparison. Then the 3D filtering and the 4NN methods are appliedto the realistic head model. The performance of different methods is evaluatedin two ways: 1) the deviation between the interpolated and the actual potentialmappings, and 2)the deviation between the re-localized and the actual dipolesource.